Most AI readiness checklists focus on data and infrastructure. What we see inside organisations is that the weakest link is almost always adoption. Here's how to run an honest assessment yourself.
Every organisation we've sat down with has the same starting position. Someone senior has a folder of AI ideas. A few of them have already run a pilot. Almost none of them have a shared, evidenced answer to a simple question: are we actually ready for this, or are we about to spend six months finding out the hard way?
An AI readiness assessment answers that question before the spending starts. It's a structured evaluation of whether your organisation has the strategic clarity, data, people, governance, and technical leadership to get value from AI, rather than just the enthusiasm to try it.
The public assessments worth knowing about, including Cisco's AI Readiness Index, RSM's, and Eide Bailly's, all converge on a similar shape: strategy, data and infrastructure, talent and culture, and governance. That convergence is useful. It tells you the categories are not controversial. What it does not tell you is where organisations actually fall down, because a glossary definition or a six-question tool cannot see your specific business. Hackett Group's own entry on the topic is a case in point: a single paragraph with no method behind it, which is exactly the gap a proper assessment needs to fill.
The scale of the gap shows up in McKinsey's 2025 Global Survey on AI: 88 percent of organisations now report using AI in at least one business function, yet the majority are still experimenting or piloting, with only around a third saying they have begun to scale their AI programmes at all. Using a tool somewhere in the business is not the same as being ready to get value from it.
Here's the pattern worth sitting with. When we assess an organisation across strategy, data, people, governance, and execution, the lowest score is almost never the data dimension, even though data is what most guides warn you about first. It's usually human adoption readiness.
The reasoning is straightforward once you look for it. A business can have moderately messy data and still get value from a well-scoped AI use case. It cannot get value from a tool that a team quietly stops using three weeks after the training session, and that happens more often than the technology press suggests. We'd put a rough figure on it: in our experience, a majority of first AI attempts inside a mid-sized organisation get meaningfully less use six months in than they did in week one. Nobody announces this. The licence stays active. The dashboard shows a handful of logins. The project is never formally called a failure, it just fades.
A senior AI leader who has run this kind of diagnostic a dozen times learns to ask a slightly different question than the standard checklists prompt. Not "do you have clean data" but "if we handed your operations team a working AI tool tomorrow, would anyone change what they do on Monday, or would it get bookmarked and forgotten." That question surfaces adoption risk faster than any infrastructure audit.
Gartner draws a similar distinction, separating AI readiness, whether the technology can meet a specific need, from human readiness, whether the organisation and its people are set up to capture and sustain the value once it's there. Most readiness conversations focus almost entirely on the first and treat the second as an afterthought, which is backwards given where the actual risk sits.
If you're going to run this internally, measure across five dimensions rather than one. Readiness tends to fail at whichever point is weakest, and a strong score in four areas will not rescue a business from a collapse in the fifth.
Strategic clarity. Does the business know where AI, data, or plain automation would create value, and has it agreed what to solve first? A long list of ideas with no ranking is not a strategy.
Data and systems readiness. Can your systems actually share what they hold, and are your key workflows documented well enough that someone outside the team could understand and improve them? This is the dimension most guides treat as the whole assessment, and it's important, but it's one of five.
Human adoption readiness. Do people feel equipped and safe using new tools, and can their managers explain what will actually change about their day? This is the dimension we'd weight most heavily, for the reason above.
Governance and risk. Is there a clear rule for what data can go into a tool, who approves a new one, and which decisions still need a human to sign off? Most organisations we see have no rule at all, which means the rule gets set informally by whoever is boldest.
Execution and technical leadership. Is there someone who can turn a business priority into a sound technical decision and stay with it through to adoption, rather than handing it to IT and hoping? This is often the missing piece in organisations between 50 and 500 people, who are too large to wing it and not yet large enough to justify a full-time Chief AI Officer.
You do not need outside help to get a first, honest picture. Here's a version you can run internally over one or two weeks.
Start by writing down, in one sentence each, what you're trying to solve. Not "we should use AI more," but the actual business problem: a reporting process that takes two days each month, a recruitment step that involves manually copying information between two systems, a marketing team spending a full day a week on first drafts. Score each candidate on how painful it is and how feasible it looks, and be honest that most of the value sits in problems this small and specific, not in an ambitious chatbot.
Then interview, don't survey. A five-question form will tell you what people think they should say about AI. A fifteen-minute conversation with someone who actually does the work will tell you where the process really breaks, and whether they trust that a new tool will still be supported next quarter.
Look at your data and workflows with a specific test in mind: could someone outside the team who owns this process understand it well enough to improve it from the documentation alone? If the answer is no, that's your data and systems score, regardless of how modern your platform is.
Write down your actual rule for AI tool approval and data handling. If there isn't one, that's the finding. Don't wait to build a full governance framework before naming the gap.
Finally, ask who in the business could take an AI priority from idea to adopted practice and stay accountable for it. If the honest answer is nobody, that's your execution gap, and it's worth naming before you commit budget to anything else.
An honest internal exercise like the one above will get you further than most organisations bother to go. What it can't easily do is see itself from the outside. People inside a business tend to rate their own strategic clarity higher than an outside interviewer would, for the same reason nobody rates their own driving as below average. And a self-assessment can tell you that adoption is weak without being able to say why, because the people closest to a problem are often the last to name its real cause.
That's the difference between a self-assessment and a structured AI readiness and value unlock audit carried out by someone with no stake in the answer. It's also why we built the AI Readiness Diagnostic as a free, five-minute version of this thinking: it uses the same five dimensions, and it's a reasonable place to start if you want a scored view before deciding whether to go further.
How long does an AI readiness assessment take?
A self-run version like the one above takes one to two weeks alongside normal work. A structured audit carried out by an outside team, including interviews and a review of systems, typically runs a few weeks depending on how many functions are in scope.
Do we need clean data before we start?
No. Data quality is one of five dimensions, not a precondition. Some of the most valuable early opportunities sit in workflows with unglamorous, moderately messy data, not in datasets that are already perfect.
What's the difference between a readiness assessment and an audit?
A readiness assessment, especially a self-assessment or free diagnostic, gives you a quick, scored view based on what you already know. An audit is carried out by an outside team, draws on interviews and a review of your actual systems, and ends with recommendations specific to your business rather than general guidance.
Is a low adoption score a people problem or a leadership problem?
Almost always leadership, even when it shows up as people quietly not using a tool. Adoption tends to fail when managers can't explain what's changing or why, not because employees are resistant to new tools in general.
What should we do with the results?
Act on the smallest, clearest opportunity first, not the largest. A process that's well understood and modestly painful is a better first move than an ambitious use case nobody has fully scoped, because it builds the internal evidence and trust that later, harder decisions depend on.
Ready to move from reading about AI to getting your organisation ready for it? Start with the diagnostic.
Take the AI Readiness Diagnostic +